1 Voltage Gated Ions Sieving Through 2D MXene Ti3C2Tx

Subscriber access provided by UNIV OF DURHAM

Article

Voltage Gated Ions Sieving Through 2D MXene Ti3C2Tx Membranes Chang E. Ren, Mohamed Alhabeb, Bryan W. Byles, Meng-Qiang Zhao, Babak Anasori, Ekaterina Pomerantseva, Khaled A. Mahmoud, and Yury Gogotsi ACS Appl. Nano Mater., Just Accepted Manuscript • DOI: 10.1021/acsanm.8b00762 • Publication Date (Web): 21 Jun 2018 Downloaded from http://pubs.acs.org on June 24, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Nano Materials

Voltage Gated Ions Sieving Through 2D MXene Ti3C2Tx Membranes Chang E. Ren1, Mohamed Alhabeb1, Bryan W. Byles2, Meng-Qiang Zhao1, Babak Anasori1, Ekaterina Pomerantseva2, Khaled A. Mahmoud3, Yury Gogotsi1* 1

A. J. Drexel Nanomaterials Institute and Department of Materials Science and Engineering

Drexel University, Philadelphia, PA 19104, USA 2

Department of Materials Science and Engineering, Drexel University, Philadelphia, PA 19104,

USA 3

Qatar Environment and Energy Research Institute, Hamad Bin Khalifa University, Qatar

Foundation, PO Box 34110, Doha, Qatar

E-mail: [email protected]

Keywords: MXene, membrane, 2D material, ion rejection, voltage gating Abstract: We report a nanochanneled Ti3C2Tx MXene membrane that enables an efficient controlled rejection/permeation of inorganic ions and organic dye molecules under applied electrical potential. When a negative electrical potential (-0.6 V) is applied to the Ti3C2Tx MXene membrane under only osmotic pressure, the rejection of inorganic salt (NaCl or MgSO4) through the membrane is enhanced. In contrast, applying a positive potential inhibits the rejection through electrostatic repulsion between charged MXene layers and inorganic cations in solution. The rejection rates of Ti3C2Tx membranes are also tested in a flow-through system. MXene membranes as thin as 100 nm show a high rejection rate above 97.9± 1.0 % for methylene blue dye molecules. Similar to inorganic salts, application of negative or positive voltages increases and decreases, respectively, the transport of molecules through Ti3C2Tx membranes. The voltage gated rejection through electronically conductive membranes is demonstrated as a promising alternative to improve rejection of inorganic salts and organic molecules.

ACS Paragon Plus Environment

1

ACS Applied Nano Materials 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 23

1. Introduction Scarcity of fresh water resources has become a critical issue that causes worldwide concern, prompting the urgent need to develop new ways to supply clean water1-2. Two main impurities in water are inorganic ions and organic molecules, and major challenges in current water treatment techniques include energy and cost effectiveness. Processes that improve the selectivity of current membranes towards impurity ions and molecules, such as microfiltration, ultrafiltration, nanofiltration and reverse osmosis, are promising approaches for decreasing energy costs among the state-of-the-art water treatment technologies3-4. However, the performance of these separation membranes is usually hindered by issues like bacterial fouling and low rejection rates to small metal ions5. In order to improve the rejection of ions/molecules, their permeation behavior through membranes must be fully understood. Optimal membranes should exhibit selective rejection of hydrated ions and particulates based on size exclusion while achieving high water flux2. For example, voltage-gated rejection of ions and molecules through electronically conductive membranes has been introduced as a smart alternative to improve membrane selectivity, rejection rate, and stability. In 1982, Murray and Burgmayer were the first to show that ion rejection can be electrically controlled using conducting polymer deposited on a gold mini-grid electrode6. Over past decades, several materials, including carbon nanotubes7, porous carbon8, metal nanotubes9-10, metal coated polymer11, and a metal-organic framework12 have shown compelling gating effect towards molecules under applied potential. One type of electroactive membrane utilizes the electrostatic interaction between the membrane and ions to gate the rejection under positive/negative voltages9, 13. In the classic model, ions in solution are electrostatically attracted to a charged surface while co-ions are repelled, creating a region where the potential decays exponentially with a characteristic length known as the Debye length of the electrical double layers (EDLs)8. Ion transport can be controlled within the Debye length using either a surface charge or electric field14-17. For porous membranes, when nanochannels are small enough that the EDLs of the channel walls overlap, direct electrostatic manipulation of ions across the nanochannel becomes possible8, 18. Another approach uses the oxidation and reduction of the membranes11, 19 or solvents20 to gate rejection. For example, a polypyrrole-based membrane was reduced by extra negative charges or oxidized by positive charges under an external electric field11,

21

. Under negative/positive potentials, the charge


2

Page 3 of 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


balance of the membrane was compensated via the cation adsorption from feed side or cation desorption to permeate side. Therefore, application of pulsed potential waveforms to the membrane resulted in metal cations flux, and the membrane was impermeable when no potential was applied. However, in most of the above cases, the feed solutions were limited to large organic molecules or coordination complexes, including methylene blue (MB), methyl orange, 2,6naphthalene disulfonate (NDS2-), p-toluene sulfonate (PTS-), Ru(bpy)32+, and [Fe(CN)6]4−8-9, 13, 22. Little electrical gating work has been focused on the rejection of metal ions. Meanwhile, twodimensional (2D) materials have been reported as novel and high-performance building blocks for separation membranes due to their efficient rejection based on steric effects in their 2D nanochannels23-24. Yet, there is little information on the electrical control of ion rejection using 2D membranes, with the exception of one study reporting a graphene-based electrochemical filtration by Zhi Zhou’s group22, which utilized the electrochemical oxidation of feed solutions to increase rejection rates. MXenes are 2D layered metal carbides and nitrides, having the general formula of Mn+1XnTx (n=1, 2, or 3), where M represents one or more transition metal elements (for example, Ti, Zr, V, Nb, Ta, Cr, Mo, etc.), X represents C and/or N, and Tx represents the surface functional groups such as (–OH, –O, or –F)25. Since their discovery, MXene family has been growing in number and attracting researchers’ attention due to the unique properties in various applications including energy storage26-27, water purification28, electromagnetic interference shielding29, optoelectronics30, and polymer nanocomposites31. MXene flakes have high aspect ratios (micrometers width to nanometer thickness), which lead to highly flexible and mechanically strong membranes31-32. The abundant surface functional groups make MXene membranes hydrophilic and stable in aqueous environments. Among more than 20 different MXene compositions synthesized to date, 2D titanium carbide Ti3C2Tx MXene is the most studied phase33. In 2015, the first layered Ti3C2Tx MXene membrane demonstrated great promise for both size and charge selective rejection of ions34. MXene composite membranes have demonstrated high rejection performance by controlling the interlayer spacing and pore slit size35-36. Ti3C2Tx MXene has negatively charged surface (-39.5 mV ξ-potential)31 and its interlayer spacing is tunable by ions intercalation with37-38 or without3940

electrical voltages. As a result of these properties, the Ti3C2Tx membranes demonstrated


3


Page 4 of 23

selectivity towards metal ions of different charge and size without external voltage

34, 41-43

. The

MXene membranes showed this effect at a much higher flux and water permeability compared to graphene-based membranes43 due to 2-3 layers of water molecules in the interlayer spacing of MXene under equilibrium in solution44-45, a significant advantage of MXenes in applications involving the removal of ions and molecules from water. Furthermore, MXene membranes are highly electronically conductive46 and have demonstrated stable performance in aqueous environments when used as electrodes for supercapacitors47 and capacitive deionization48. Thus, it is highly feasible to control Ti3C2Tx membranes’ rejection rates towards metal ions via the application of external voltages. To demonstrate this idea, in this study, electrical potential (both positive and negative) was applied on 2D Ti3C2Tx MXene membranes to evaluate how external potential would affect the rejection rates of inorganic metal ions like Na+ and Mg2+ as well as organic MB+ ions. The advantage of the proposed strategy is the ability to enhance or inhibit ion rejection via application of voltage to control intercalated ions and tune the interlayer spacing of the Ti3C2Tx. 2. Results and Discussion The Ti3C2Tx membranes used here are prepared by a vacuum-assisted filtration method. The cross-sectional scanning electron microscopy (SEM) image in Figure 1a shows orderly stacked 2D nanochannels across the structure of a Ti3C2Tx membrane, and the bottom inset is a typical top-view digital image of a 4-cm-diameter membrane (gray circle in Figure 1a inset). The Ti3C2Tx membrane had a thickness of ~1.5 µm and was attached on a PVDF filter paper (pore size 450 nm) for mechanical support. The Ti3C2Tx flakes used here were produced by the minimally intensive layers delamination (MILD) route developed recently46,

49

, with manual

shaking instead of sonication to avoid breaking the MXene ﬂakes into smaller pieces. The resulting MXene flakes made via the MILD method showed larger lateral size of 3 to 15 µm with cleaner edges49-50 compared to the previously reported small flakes (< 1 µm in lateral size) obtained by sonication51. The electrical conductivity of the membrane composed of large-size Ti3C2Tx flakes was determined to be 5490± 140 S/cm using a standard four-probe technique, higher than that of a small-size Ti3C2Tx flake membrane (4540± 260 S/cm). In addition, membranes that are made from larger Ti3C2Tx flakes have better mechanical stability and flexibility due to larger lateral size as well as better connection between cleaner flakes. Thus, throughout this study, we focus on the Ti3C2Tx membranes made from larger lateral-size flakes


4

Page 5 of 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


produced via the MILD route with only manual hand shake process, labeled as larger-flake membrane. To gain insights on the effect of flake size, we also produce Ti3C2Tx membranes from a Ti3C2Tx colloidal solution bath sonicated for 1.5 h, labeled as smaller-flake membrane, and compare the results to the larger-flake membrane. It was mentioned earlier that ion transport can be electrostatically manipulated within the Debye length of the EDLs using either surface charge or electric field. The thickness of the EDL, or the Debye length, λD, in an aqueous solution containing a symmetric monovalent salt, such as sodium chloride (NaCl), can be calculated using equation (1)52: =

(1),

where ε0 is the permittivity of free space, εr is the dielectric constant of the solution, R is the gas constant, T is the absolute temperature in kelvins, F is the Faraday constant, and C0 is the molar concentration of the electrolyte. For NaCl solution with a concentration of 35 g/L, λD is calculated to be ~0.4 nm. Thus, to form two individual EDLs between two adjacent Ti3C2Tx layers, an interlayer distance of at least 0.8 nm is required. Meanwhile, the nanochannel height for wet Ti3C2Tx membranes was measured to be ~0.29 to 0.64 nm34, 53 depending on the number of water molecules at the MXenes interlayer galleries. Note that we consider the channel between two MXene sheets as the interlayer spacing in this study as shown in Figure 1b39, 54. Since the 2D nanochannel height is only ~0.6 nm, the EDLs overlap, as shown in Figure 1a. This means that after intercalation into the nanochannels of the MXene membrane, ions enter a space that is fully under the control of MXene surface charges. Thus, controlling the ion rejection by exerting external voltage and changing Ti3C2Tx surface charge is theoretically feasible. It is known that the salt solutions need to maintain electrical neutrality, and therefore, the salt cations and anions must be transported in a stoichiometric ratio from feed to permeate solution55-57. Atomic absorption spectroscopy (AAS) was used to quantify the amount of Na+ and Cl- ions in the permeated solution (Figure S1) and revealed the Na:Cl ratio was approximately 1:1 in the permeate solution, confirming ions were rejected in a stoichiometric 1:1 ratio. The assumed behavior of anions is that they are dragged by cations to permeation, after cations intercalated and permeated through the Ti3C2Tx membrane, due to the electrostatic attraction between cations and anions and the need to maintain electrical neutrality.


5


Page 6 of 23

To investigate the ion rejection under only osmotic pressure without external vacuum pressure, a U-shaped device (Figure 2a), was used to measure the number of permeated ions over a given time. NaCl solution of 35 g/L (approximate NaCl concentration in seawater) was used as the major feed solution. 50 mL of NaCl solution was introduced into the feed side (left side in Figure 2a), and 50 mL of DI water was introduced into the permeate side (right side in Figure 2a). For the electrical control, a platinum wire was used as the counter electrode and Ag/AgCl in 1M KCl as the reference electrode (shown as RE in Figure 2a), with both electrodes immersed in the feed solution. The Ti3C2Tx membrane on the PVDF support was attached to an annular Ti foil, which served as the current collector (CC), and pressed between feed and permeate solution containers. The voltages were applied by connecting to an electrochemical workstation (VMP3, Biologic, France). A real-time ionic conductivity meter was immersed in the permeate side to record the ionic conductivity, with a Teflon stir bar constantly stirring the solution to avoid a gradient of concentration.

Figure 1. a) Cross-sectional SEM image of a Ti3C2Tx membrane (the inset shows a digital top-view image of a 1.48 µm-thick Ti3C2Tx membrane on supporting PVDF filter paper), with a schematic showing the interaction between Ti3C2Tx layers and ions within their EDLs when the Ti3C2Tx membrane is connected to a negative voltage. b) A molecular model of Ti3C2Tx showing two layers of water with ions between the layers. We consider the channel between MXene flakes as the interlayer spacing. As a result, c lattice parameter (c-LP) calculated from XRD is two Ti3C2Tx flake thicknesses and two interlayer spacings.


6

Page 7 of 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


The ionic conductivity on the permeate side was recorded at a rate of 1 point/s immediately after the permeation started. The conductivity of the permeate side was converted to the concentration of salt ions, and the number of permeated ions was calculated using the classical diffusion equation as described in the experimental section. Figure 2b shows the number of permeated cations in mol/m2 under different voltages as a function of the permeation time for the larger-flake Ti3C2Tx membrane. The slopes of the curves determine how fast the ions permeated through the Ti3C2Tx membrane, or the permeation rates. At the beginning, no external voltage was applied, and the number of permeated ions increased linearly at a permeation rate of ~ 0.122 mol/(m2 h). After 30 min, when a steady flow rate is achieved, positive voltage (0.4 V) was applied to the MXene membrane, and the number of permeated ions exhibited quadratic growth and then grew linearly at a rate of 0.260 mol/(m2 h). This permeation rate is higher than the rate under no applied voltage, indicating the positive potential leads to acceleration of ion transport. Notably, when the voltage was then changed to a negative value of -0.6 V for 30 min, the permeation rate dropped to 0.003 mol/(m2 h), suggesting effective blocking of ion transport. The blocking effect shown here was demonstrated at a lower potential (-0.6 V) than a similar study on mesoporous carbon that required -0.9 V vs. Ag/AgCl to for reduced transport of KCl through the membrane8. This is likely due to the smaller controllable interlayer spacing of the MXene (~0.6 nm) compared to the ~7.8 nm pore size in the carbon membrane. Another cyclic change of the applied external voltages (0.4 and -0.6 V) was conducted and the permeation rates changed with the same trend but of lesser magnitude. The permeation rates under different potentials are listed in Table 1. Magnesium sulfate (MgSO4) with a concentration of 3.4 g/L (approximate MgSO4 concentration in seawater) was also used as the feed solution. A similar relationship was found between the permeation rates and applied voltages, as is shown by the bottom red curve in Figure 2b and its expanded view in the inset. MgSO4 permeation rates were generally much lower (Figure 2b inset), which is attributed to the larger hydrated size and higher charges of Mg2+ and SO42- ions compared to Na+ and Cl- ions, the lower molar concentration of MgSO4 solution, and the intercalated Mg2+ ions that are more attracted to negatively charged MXene layers38.


7


Page 8 of 23

Table 1 The NaCl salt permeation rates through Ti3C2Tx membrane under different potentials in one cyclic change*

Time (h)

0 – 0.5

0.5 – 1

1 – 1.5

1.5 – 2

2 – 2.5

2.5 – 3

Potential (V)

0

0.4

-0.6

0.4

-0.6

0

0.260**

0.003

0.223

0.030

0.068

Permeation rate 0.122 (mol/(m2 h)) *

The rates were calculated from the slope of permeation curves in each 30-min cycle in Figure 2b. The permeation rate when 0.4 V was applied for the first time was the slope of the permeation curve between the time 0.75 – 1 h. **

Cyclic applications of no potential, negative potential, and positive potential in NaCl were repeated four times over 6 hours to check the stability of the voltage controlling effect (Figure 2c). The ion blocking effect at negative voltages and ion permeation accelerating effect at positive voltages were maintained throughout the test, though after several cycles the effects became less pronounced compared to the first cycle. This damping may be caused by the residual ions between Ti3C2Tx layers or the surface charge residues after cycling, similar to the less pronounced interlayer spacing change in MXene electrodes during electrochemical cycling58. Further, swelling of electrodes due to water and salt penetration between the layers is a typical problem for 2D materials use in desalination/filtration membranes59. For a control test, blank PVDF was used as the separation membrane with cyclic voltages applied, and no reversible changes on the permeation rate were observed (Figure S3). To ensure the Ti3C2Tx electrode was chemically stable and no hydrogen or oxygen evolution reaction occurred under the applied voltages, the Ti3C2Tx electrode was tested in a range covering the applied voltages using a three-electrode Swagelok setup with details described in SI. In Figure S4, the cyclic voltammetry profiles showed that there were no obvious reaction peaks in the voltage range of -0.6 to 0.4 V, and no adverse effects or parasitic reactions were found. In order to study the effect of the lateral size of 2D MXene flakes, we fabricated a similar thickness membrane using smaller Ti3C2Tx flakes and compared its permeation rate with the larger-flake Ti3C2Tx membrane. The number of permeated cations in experiments with NaCl


8

Page 9 of 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


solution is compared for these two membranes in Figure 2d. The smaller-flake MXene membrane demonstrated similar voltage gating effects on ion rejection as the larger-flake MXene membrane, but the magnitude of the permeation rates was higher for the smaller-flake MXene membrane. Similar phenomenon was previously observed for graphene oxide membranes with different flake sizes60. The higher permeation rates of the membrane with smaller flakes were attributed to the shortened ion transport pathways around the smaller flakes. In MXene membranes, ions and water molecules are transported along a zig-zag pathway around the orderly stacked Ti3C2Tx flakes. Larger Ti3C2Tx flakes reduce the number of flakes and openings between flakes in a given membrane, resulting in longer and more tortuous ion transport pathways for ions and thus slower permeation. Because the larger-flake Ti3C2Tx membranes are more mechanically stable, more electrically conductive, and show slower permeation rates compared to smaller-flake membranes, we focus on the former for the rest of this study.

Figure 2. a) A schematic showing the U-shaped device that was used to measure the ion permeation through Ti3C2Tx membrane under external potentials, with the platinum wire as a counter electrode (CE), Ag/AgCl in 1M KCl as a reference electrode (RE), a Ti3C2Tx membrane on PVDF support as a working electrode (WE), and an annular Ti foil as a current collector (CC). The number of permeated cations as a


9


Page 10 of 23

function of time and applied voltages of b) NaCl and MgSO4 through Ti3C2Tx membrane (larger flakes) for 3 h (the inset shows the expanded view of the curve), c) NaCl through Ti3C2Tx membrane (larger flakes) for 6 h, and d) NaCl through Ti3C2Tx membrane of larger and smaller flakes for 3 h. The concentrations of the NaCl solution MgSO4 solutions were 35 g/L and 3.4 g/L, respectively.

Cation intercalation is a common behavior for Ti3C2Tx films with or without external voltages51, 53-54, and it causes changes in interlayer spacing37-38. We used X-ray diffraction (XRD) to observe the changes of interlayer spacing caused by ion intercalation. To do so, we stopped the test at three different voltages of 0, 0.4, and -0.6 V after permeation with NaCl solution, dried the membranes under vacuum at room temperature, and performed XRD measurements on each membrane film. The XRD patterns of these membranes are compared with that of the vacuum dried as-prepared Ti3C2Tx membranes (bottom pattern) in Figure 3. The c-lattice parameter (c-LP) can be calculated from the position of the (0002) peaks of Ti3C2Tx according to the Bragg’s law, which correspond to the thickness of two Ti3C2Tx layers plus two effective interlayer spacings as shown in Figure 1b39. Because the (0002) peak usually has the highest intensity, it is used to measure the c-LPs of MXenes26. One rigid layer of Ti3C2Tx has a thickness of 8.8 Å from MD simulations, thus the interlayer spacing (IS) can be calculated. As shown in Figure 3, the assynthesized Ti3C2Tx membrane had the smallest IS of 3.7 Å. After intercalation at 0 V, the IS shows an increase to 5.8 Å, possibly due to Na+ ions intercalation into the membrane. For the membrane intercalated at 0.4V positive voltage, the IS was further increased to 6.5 Å, due to the repulsion of intercalated Na+ and positive MXene surfaces. In the membrane after tested at a negative voltage of -0.6 V, the IS returned to 4.1 Å (top pattern, Figure 3), close to its original value. This was ascribed to an efficient attraction of Na+ to negatively charged MXene and lead to the lowest permeation rates of Na+. The changes in MXene interlayer spacing in the membranes were affected by both the size of inserted cations and electrostatic interaction between cations and Ti3C2Tx layers. The insertion of cations with solvation shells expands the MXene interlayer spacing, while the electrostatic attraction between cations, especially multivalent ones38, and negatively charged MXene layers contracts the interlayer spacing34, 38. The smallest increase of interlayer spacing under negative voltage agrees well with the contraction due to electrostatic attraction, and the resulting constrained and limited nanochannels are the reason for hindered salt permeation. On the contrary, under positive voltage, the intercalated cations opened up the interlayer by repelling


10

Page 11 of 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


with positively charged Ti3C2Tx surfaces, leading to the largest interlayer expansion and increased permeation rate compared to the situations under neutral and negative voltages. It is worth mentioning that the opposite trend in the MXene interlayer spacing change under applied voltages was observed in ionic liquid, with interlayer expansion under negative voltages, and contraction happening under positive voltages58, 61. Compared to metal ions, the ions of ionic liquid have larger size, and the size/charge ratio is much larger. The expansion due to size accommodation is much larger than electrostatic contraction. When applying negative voltages, more cations of ionic liquid intercalated and caused expansion; while under positive voltages, the deintercalation of cations or attraction between positive MXene and anions caused contraction.

Figure 3. The X-ray diffraction patterns of Ti3C2Tx membranes (pristine and permeated with NaCl salts at 0, 0.4, and -0.6 V). In all cases, the interlayer spacings of Ti3C2Tx layers under different conditions are noted.

In industrial applications, the rejection rate of ions or molecules is measured with water flowing through membranes. In this consideration, a dead-end filtration setup consisting of a 1.5-


11


Page 12 of 23

µm-thick Ti3C2Tx membrane (larger-flakes) attached on a PVDF membrane was used to measure the rejection rates of solutions (Figure 4a). Inorganic salt solutions (NaCl, AlCl3, etc.) and MB dye solution (10 ppm) were used as the feed. The solution was transported through the membrane under a vacuum pressure of 1 bar. By comparing the concentration of feed and permeate, the rejection of inorganic salt was found to be as low as 10 to 20 % and was not increased under external voltage. The low rejection may be due to the small size of ions, fast water flow, and swelled interlayer spacing of Ti3C2Tx, as previously observed for GO membranes59. To measure the rejection of MB, the concentrations of feed and permeate MB were measured by UV-vis spectroscopy, as shown in Figure 4b. By comparing the absorbance at peak of MB solutions, the rejection rates to MB through Ti3C2Tx membranes were calculated and shown in Table 2. To verify the required thickness for the membrane to reject MB, we fabricated Ti3C2Tx membranes as thin as ~100 nm, which showed a similar rejection rate as the 1.5-µmthick membrane of ~98.0 % and notably, a much higher water flux (Table 2). This implied that the Ti3C2Tx film membranes as thin as 100 nm can be pinhole-free, possibly due to the highly regular stacking of Ti3C2Tx flakes, as shown earlier in Figure 1a. This thickness is less than those of polymer cellulose and polysulfone membranes62-63 by two orders of magnitude, which showed similar rejection rate of MB. Motivated by the previously reported superior mechanical stability of MXene/polymer membranes31, Ti3C2Tx/polyvinyl alcohol (PVA, 10 wt.%) membranes were also fabricated and tested for extended cycling. The Ti3C2Tx/PVA membranes demonstrated rejection rates of MB almost as high as the pure Ti3C2Tx (97.2± 0.8% for Ti3C2Tx/PVA vs 98.0± 0.9 % for Ti3C2Tx). It is worth noting that though mechanically stable under testing conditions, pure Ti3C2Tx membranes could not be retested after drying due to membrane breakage, while the Ti3C2Tx/PVA membranes could be reused multiple times without obvious performance degradation. The improved stability is attributed to the strong interaction between Ti3C2Tx and PVA by hydrogen bonds. Voltages were applied on the Ti3C2Tx/PVA membranes to tune the permeation rate. The filtration set-up for these experiments is shown in Figure 4a: an annular Ti foil on the top worked as counter electrode, a rubber layer worked as an insulter in the middle, and the Ti3C2Tx membrane as working electrode was attached to current collector (Ti foil). Under external voltages, Ti3C2Tx/PVA membranes’ rejection was further enhanced to 99.6± 0.3 % by applying a negative voltage of -0.5 V, and the rejection was depressed to 92.5± 1.1 % under a


12

Page 13 of 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


positive voltage of 0.1 V. The electrostatic manipulation of dye molecules rejection was due to the control of MXene nanochannels under potentials as well.

Figure 4. a) A schematic showing the dead-end filtration device used to measure the rejection rates through Ti3C2Tx membrane with applied voltages. b) The UV-vis curves of feed MB and permeated MB solutions through Ti3C2Tx membrane under different conditions, including the MB solution before permeation, and filtered MB solutions through Ti3C2Tx membranes of 1500 and 100 nm thickness under 0V, and through Ti3C2Tx/PVA of 1500 nm thickness under 0, -0.5 and 0.1 V applied voltage.

Table 2. The rejection rates of MB and NaCl and flux through Ti3C2Tx, polysulfone and GO/PVA membranes with different thicknesses under potentials

Thickness (nm)

Voltage (V)

Rejection Flux (L/m2 rate (%)* bar h)

1500

0

98.0± 0.9

10.6

100

0

97.9± 1.0

10,498

1500

0

97.2± 0.8

7.4

1500

-0.5

99.6± 0.3

5.6

1500

0.1

92.5± 1.1

60.1

Polysulfone62-63

200,000

0

98.5- 99.4 ~100

GO/PVA64

120

0

99.8 (NaCl) 18.9

Membrane Ti3C2Tx

Ti3C2Tx/PVA

*

The rejection rate is of MB if not denoted otherwise.


13


Page 14 of 23

In addition to the voltage gated ion rejection performance, Ti3C2Tx membranes have another advantage as separation membranes, which is their anti-fouling potential. Ti3C2Tx have shown bactericidal properties against Gram (−) E. coli and Gram (+) B. subtilis bacteria65. In addition, Vecitis’ group showed that positive voltage on carbon nanotubes membranes reduced the number of bacteria and viruses in the permeate to below the detection limit66. Similar behavior is expected for MXene membranes that exerting voltage can enhance bacterial removal, and this has yet to be explored. Further, this report establishes a new area of research for MXene membranes. Although electrically controlled ion rejection is promising due to the advantages of lower energy and pretreatment demand, it faces several key challenges, including the scalability of the technology to a commercially viable level and low salt rejection rates so far. It is expected that the promising control of interlayer spacing and rejection of salts shown here can be improved and maintained with extended application by limiting the swelling of the nanochannels from water flow. This swelling of the nanochannels leads to increased layers of water between MXene layers, and the resulting expanded interlayer distance causes a decrease in performance after repeated cycles of applied potential or under high water flux conditions. Two promising approaches to prevent this swelling of MXene membranes and maintain the voltage gating effect are physically confining the MXene membrane67 via mechanical compression or introducing chemical bonding between MXene layers31. Moreover, the onset potential value at which control of rejection is achieved must be determined to obtain rejection control with the lowest energy consumption. 3. Conclusions This work showed effective control on the rejection of inorganic salts (NaCl or MgSO4) and organic methylene blue molecules through 2D Ti3C2Tx MXene membranes by applying an electrical potential. Negative and positive voltages increased and decreased ion rejection through MXene membranes, respectively. Ti3C2Tx MXene membranes as thin as 100 nm tested under vacuum pressure showed high rejection over 97.9± 1.0 % to methylene blue dye solution, along with high water flux of 10,498 L/m2 bar h. Further, application of external voltages increased/decreased the rejection of methylene blue dye based on the charges. The high water flux and promising controllable rejection behavior demonstrated here highlight the advantages of MXene membranes for electrically controlled ion rejection. Mechanical compression and


14

Page 15 of 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


bonding between MXene sheets are promising pathways for preventing swelling and maintaining the controllable rejection. Additionally, removal of ions from the MXene membranes with electrical voltage offers a potentially effective way to regenerate the membranes and elongate their lifetime. This unique behavior of the Ti3C2Tx MXene membranes opens new horizons in controlled ion selectivity, desalination, and chemical separations. 4. Experimental Section Preparation of Ti3C2Tx-based membranes Ti3AlC2 MAX powder (

1 Voltage Gated Ions Sieving Through 2D MXene Ti3C2Tx

Recommend Documents